Electronic Appendix
MRI methodology
The implantable coil consisted of a single-tuned phosphorus (31P) match+tune+tune-balance variable capacitor network using a Helmholtz-shaped transmit-receive coil, manufactured at the Howard Florey Institute. In situ tuning was achieved using a frequency sweep generator (Wavetek, CA, USA; model 1062).
To quantify ATP, 31P MR spectra were recorded every 15 minutes from the implanted coil and acquired at 51.705 MHz in a 3.0 Tesla magnetic resonance scanner (LX horizon, GE Healthcare, WI) using a flip angle of 90°and a relaxation delay of 2 sec. Spectra were recorded from a 10 cm thick slice using a spin-echo sequence, with 2048 data points acquired over a spectral width of 16000 Hz and 450 transients averaged per spectrum to produce a total acquisition time of 15 min.
To determine the turnover rate for the terminal phosphate through ATPase (i.e. from ATP to Pi), the saturation transfer technique [13] was employed. Of the two spectra acquired for each saturation transfer experiment, one had the saturation pulse centred on ATP and the other had the saturation pulse centred on 2.5 ppm. The two spectra were then subtracted and the change in the (intracellular) inorganic phosphate Pi peak area was determined. This was combined with the calculated Pi and ATP concentrations to estimate the rate of phosphate turnover through ATPase. Raw data were zero filled twice and a 10 Hz Lorentzian apodisation function was applied prior to Fourier Transformation. The small peak at higher frequency adjacent to the ATP signal was assigned to phosphocreatine and referenced to 0.0 ppm.
Peak areas under the curve were determined using the integration routine of the SAGE program (GE Healthcare, Fairfield, CT). Total phosphate was calculated as the sum of the areas under the curves for phosphodiesters, phosphomonoesters, inorganic phosphate, ATPα, ATP and ATP. Total ATP was calculated as the sum of ATPα, ATP and ATP. Data for total, α, and ATP are presented as a fraction of total phosphate.